Cerebrospinal fluid biomarkers for predicting development of multiple sclerosis in acute optic neuritis: a population-based prospective cohort study

BACKGROUND: Long-term outcome in multiple sclerosis (MS) depends on early treatment. In patients with acute optic neuritis (ON), an early inflammatory event, we investigated markers in cerebrospinal fluid (CSF), which may predict a diagnosis of MS. METHODS: Forty patients with acute ON were recruited in a prospective population-based cohort with median 29 months (range 19–41) of follow-up. Paired CSF and serum samples were taken within 14 days (range 2–38), prior to treatment. Prospectively, 16/40 patients were by a uniform algorithm diagnosed with MS (MS-ON) and 24 patients continued to manifest isolated ON (ION) during follow-up. Levels of cytokines and neurofilament light chain (NF-L) were measured at the onset of acute ON and compared to healthy controls (HC). Significance levels were corrected for multiple comparisons (“q”). The predictive value of biomarkers was determined with multivariable prediction models using nomograms. RESULTS: CSF TNF-α, IL-10, and CXCL13 levels were increased in MS-ON compared to those in ION patients (q = 0.021, 0.004, and 0.0006, respectively). MS-ON patients had increased CSF pleocytosis, IgG indices, and oligoclonal bands (OCBs) compared to ION (q = 0.0007, q = 0.0058, and q = 0.0021, respectively). CSF levels of IL-10, TNF-a, IL-17A, and CXCL13 in MS-ON patients correlated with leukocyte counts (r > 0.69 and p < 0.002) and IgG index (r > 0.55, p < 0.037). CSF NF-L levels were increased in ON patients compared to those in HC (q = 0.0077). In MS-ON, a progressive increase in NF-L levels was observed at 7 to 14 days after disease onset (r = 0.73, p < 0.0065). Receiver-operating characteristic (ROC) curves for two multivariable prediction models were generated, with IL-10, CXCL13, and NF-L in one (“candidate”) and IgG index, OCB, and leukocytes in another (“routine”). Area under the curve was 0.89 [95% CI 0.77–1] and 0.86 [0.74–0.98], respectively. Predictions of the risk of MS diagnosis were illustrated by two nomograms. CONCLUSIONS: CSF TNF-α, IL-10, CXCL13, and NF-L levels were associated with the development of MS, suggesting that the inflammatory and neurodegenerative processes occurred early. Based on subsequent diagnosis, we observed a high predictive value of routine and candidate biomarkers in CSF for the development of MS in acute ON. The nomogram predictions may be useful in the diagnostic work-up of MS. ELECTRONIC SUPPLEMENTARY MATERIAL: The online version of this article (10.1186/s12974-019-1440-5) contains supplementary material, which is available to authorized users.